A Brief Introduction to Human Behavioral Pharmacology: Methods, Design Considerations and Ethics

Abstract

Human behavioral pharmacology methods have been used to rigorously evaluate the effects of a range of centrally acting drugs in humans under controlled conditions for decades. Methods like drug self-administration and drug discrimination have been adapted from nonhuman laboratory animal models. Because humans have the capacity to communicate verbally, self-report methods are also commonly used to understand drug effects. This perspective article provides an overview of these traditional human behavioral pharmacology methods and introduces some novel methodologies that have more recently been adapted for use in the field. Design (e.g., using placebo controls, testing multiple doses) and ethical (e.g., avoiding enrollment of individuals seeking treatment, determining capacity to consent) considerations that must be addressed when conducting these types of studies are also described.

Introduction

Controlled human laboratory research methods have been used to assess the behavioral and pharmacological effects of centrally acting drugs, in particular drugs of abuse, since at least the 1940s (e.g., Isbell & Wikler, 1947; Williams et al., 1946), with much early work being done at the U.S. Public Health Service Addiction Research Center in Lexington, Kentucky (Campbell, 2006). It should be noted that this work was conducted with prisoners who had been sent to the prison hospital in Lexington because of their substance-use disorders (then termed “addiction”) and this center eventually became the Intramural Research Program at the National Institute on Drug Abuse (for further information, see Campbell, 2006). Human behavioral pharmacology research brought together experts in the fields of clinical pharmacology, psychology, physiology, biochemistry, psychiatry, and biophysics (Campbell, 2010), which enabled the rigorous, multifaceted assessment needed to more fully understand the effects of drugs of abuse in humans and, in turn, substance-use disorders.

This type of research proliferated and is now conducted at numerous institutions, including the University of Kentucky, Columbia University, Brown University, Johns Hopkins University, the University of Minnesota, Virginia Commonwealth University, and the University of Chicago. This research approach has been used to determine basic biological, behavioral, and pharmacological mechanisms related to substance-use disorder, to evaluate the abuse potential of novel compounds and to screen putative substance-use disorder treatments prior to advancement to clinical trials (Comer, Ashworth, et al., 2008a; Czoty et al., 2016; de Wit & Phillips, 2012; Haney & Spealman, 2008; Romach et al., 2014; Van Hedger et al., 2017). Because this work sits between preclinical (e.g., in vitro and nonhuman animal research) and clinical trial research1 on the translational spectrum in substance-use disorder research, human behavioral pharmacology research may be considered part of a pipeline in developing and advancing novel medications and treatments (Czoty et al., 2016; Horton et al., 2013).

The purpose of this article is to provide a brief introduction to methods used in human behavioral pharmacology research. Classic and long-standing human laboratory measures like drug self-administration, drug discrimination, and subjective drug-effects questionnaires will be featured. More novel measures, like behavioral economics measures (e.g., delay discounting, drug purchase tasks) and attentional bias tasks, will also be touched upon. Lastly, practical study design considerations and ethical issues associated with conducting this type of research will be covered. Of course, because this perspective piece is only intended as an introduction, interested readers are referred to several other excellent resources, beyond those cited above, that have also covered topics relevant to human behavioral pharmacology research (College on Problems of Drug Dependence, 1995; Comer, Bickel, et al., 2010a; Foltin & Fischman, 1991; Fischman & Johanson, 1998; Griffiths et al., 20032).

Methods

Classic Methods

The most commonly used human behavioral pharmacology methods include drug self-administration, drug discrimination (but see below) and subjective drug effects questionnaires (see Table 1). Drug self-administration and drug discrimination methods in humans map closely onto those methods in nonhuman animals, adjusting for some differences due to being able to instruct humans on how to perform these tasks, whereas subjective drug effects questionnaires capitalize on humans’ unique ability to verbally report their experience following drug administration. Although subjective response to drugs has been evaluated since the earliest human behavioral pharmacology research (e.g., administration of n-allylnormorphine antagonized the euphoric effects of morphine; Wikler et al., 1953), human drug self-administration and drug discrimination methods were developed somewhat later (e.g., Altman et al., 1976; Bigelow et al., 1975a, 1975b; Hurst et al., 1973; but see O'Driscoll & Lindley, 1957, for an interesting case report).

Table 1.

Summary of classic human behavioral pharmacology methods

Measure	Brief description	Primary outcome
Self-Administration	Objective measure of willingness to engage in drug taking behavior	Drug Intake (e.g., number of drug choices) or Indicator of Responses Completed (e.g., break point on a progressive ratio)
Drug Discrimination	Objective measure of discriminative stimulus effects of a drug	Percent Drug Appropriate Responding
Subjective Effects	Self-report of interoceptive drug effects	Quantified Response to Drug Effect Question (e.g., number of units on a 100-unit scale, score on a Likert-type scale)

Self-Administration

Similar to drug self-administration methods in nonhuman animals, drug self-administration measures in humans provide an objective index of an organism’s willingness to engage in drug-taking behavior. In a typical arrangement, human subjects first sample a dose of the drug (or drugs) under study and are then offered opportunities to work to earn that dose again (for a more detailed review, see Jones & Comer, 2013). Different schedules of reinforcement have been used in human laboratory drug self-administration procedures, from simple choice (i.e., a subject verbally indicating that they would like to take a drug dose again, which can be conceptualized as a fixed ratio 1 schedule) to more complex progressive ratio schedules (Johanson & de Wit, 1989; Stoops, 2008). The outcomes are typically expressed as a single variable, such as the number of drug choices or “break point” on a progressive ratio schedule. In more recent years, choice measures (with or without a more complex schedule of responding) have become commonly used, in particular models in which human subjects make choices between a drug dose and an alternative reinforcer because such arrangements may map better onto the clinical conditions observed in people with substance-use disorders (Moeller & Stoops, 2015).

Depending on the route of administration and doses administered, subjects may be offered the opportunity to work (e.g., press a key on a keyboard, click a mouse) for the full sampled dose numerous times in a session or they may work for portions of that sampled dose, with the amount of drug earned being administered all at once after the self-administration task. For example, Foltin and Fischman (1992) allowed subjects to first sample doses of smoked (0, 25, and 50 mg) and intravenous (0, 16, and 32 mg) cocaine and then make five choices between those full-sampled doses because parenterally administered cocaine has a short half-life (Cone, 1995). For drugs with longer half-lives (e.g., d-amphetamine) or with routes that take longer to onset (e.g., oral), researchers have turned to having subjects work for portions of a sample dose. We previously evaluated the reinforcing effects of oral d-amphetamine (0, 8, 16, and 24 mg) and methylphenidate (0, 16, 32, and 48 mg; Stoops et al., 2004). In that study, subjects first sampled full doses of the drug, then could self-administer portions of that dose, with one eighth of each dose being available during each component of an eight-component progressive ratio schedule. In both of those studies, active doses of the tested drugs were self-administered to a greater degree than placebo, indicating the sensitivity of either approach to the reinforcing effects of drugs in humans while also allowing for multiple opportunities to measure drug taking behavior. In addition to being used to evaluate/compare the reinforcing effects of drugs as in the studies described above, self-administration studies have also examined how putative behavioral or pharmacological interventions change drug intake in humans (e.g., Arout et al., 2021; Lile et al., 2020; Stoops et al., 2010, 2019; Sullivan et al., 2006).

There is strong concordance between human and nonhuman self-administration findings, with most drugs that function as reinforcers in nonhuman animals doing the same in humans (Panlilio & Goldberg, 2007). Drugs that are used recreationally by humans in the natural ecology also generally function as reinforcers in the human laboratory. In both human and nonhuman studies, self-administration measures have routinely been used to determine the reinforcing effects of drugs, in particular novel compounds, as well as to compare reinforcing effects across drugs (e.g., Comer, Sullivan, et al., 2008b; Gannon et al., 2018; Hiranita et al., 2009; Stoops et al., 2004). Further, as has been shown in nonhuman animals (e.g., Nader & Woolverton, 1991, 1992), increasing the magnitude of an alternative reinforcer or increasing the response cost for a drug decreases drug intake (e.g., Foltin et al., 2016; Lile et al., 2016). Lastly, in nonhuman animals and humans, context and/or environmental influences play an important role in drug self-administration (e.g., Comer, Sullivan, et al., 2010b; Gipson et al., 2011; Stoops, Lile, Fillmore, et al., 2005a; Strickland & Smith, 2015) Unlike nonhuman studies, however, human laboratory methods do not typically rely on response rate as a measure of reinforcement and, perhaps due to limited dose ranges in humans, drug self-administration in humans is often observed as a monotonic increasing function of dose. Human laboratory self-administration methods have also come to be seen as a “screen” for interventions for substance-use disorders, with good predictive validity, whereas preclinical research may have less predictive validity (Comer, Ashworth, et al., 2008a; Czoty et al., 2016; Haney & Spealman, 2008; Venniro et al., 2020).

Drug Discrimination

As with drug-discrimination methods in nonhuman animals, drug discrimination in humans determines the discriminative stimulus effects of a centrally acting drug. In a typical arrangement, human subjects first sample a dose of the training drug, often labeled Drug A, and then enter a test of acquisition phase in which they are asked to discriminate between the training drug and placebo, often labeled “Not Drug A” (or Drug B). After meeting some discrimination accuracy criterion (e.g., 80% correct responding over four consecutive sessions), subjects then enter a test phase in which a range of doses of the training drug, other drugs, or combinations of the training drug with other drugs are evaluated (for a more detailed review, see Bolin et al., 2016). Drug-discrimination tasks in humans are set up to differentially reinforce accurate responding using a consequence like money. For example, after administration of a drug dose, subjects will repeatedly be given the opportunity to distribute points between Drug A and Not Drug A (or Drug B) options (e.g., Kamien et al., 1997; Stoops, Lile, Glaser, & Rush, 2005b). Accurately allocated points (i.e., points allocated to Drug A when the training drug is administered, points allocated to Not Drug A when placebo is administered) are exchangeable for money. When novel doses, drugs, or drug combinations are administered, the option with the most allocated points is often selected for monetary exchange. Drug discrimination outcomes are often expressed as a single variable, such as percent drug appropriate responding.

Drug-discrimination studies in humans routinely use oral dosing (but see Johanson et al., 2006; Schuh et al., 2000) and typically fall into one of two approaches: drug substitution (i.e., comparing other drugs to the training drug) or drug combination (i.e., evaluating how the effects of combining the training drug with another drug) studies. In substitution studies, after subjects successfully acquire a discrimination, the effects of novel drugs (and often novel doses of the training drug) are evaluated. For example, in one study, after subjects had learned to discriminate 10 mg oral methamphetamine from placebo, a range of oral doses of methamphetamine (2.5–15 mg), d-amphetamine (2.5–15 mg), methylphenidate (5–30 mg), and triazolam (0.0625–0.375 mg) were then tested (Sevak et al., 2009). As expected, methamphetamine, d-amphetamine, and methylphenidate all produced dose-dependent increases in methamphetamine appropriate responding whereas triazolam did not. The inclusion of triazolam as a negative control is important because it demonstrates that the discrimination was pharmacologically selective (i.e., subjects were not simply discriminating between the presence or absence of a drug). In a subsequent drug-combination study, the discriminative stimulus effects of methamphetamine were determined alone and following pretreatment with the monoamine partial agonist/antagonist, aripiprazole (20 mg; Sevak et al., 2011). Methamphetamine alone produced dose-related increases in methamphetamine appropriate responding and aripiprazole attenuated methamphetamine appropriate responding.

There is strong concordance between human and nonhuman findings in drug-discrimination measures. In particular, drugs that function as discriminative stimuli in humans also do so in nonhuman animals (Kamien et al., 1993) and drugs with similar neuropharmacological activity tend to produce similar discriminative stimulus effects (Bolin et al., 2018; Kelly et al., 2003). An important difference, though, is that instructional control allows humans to acquire a drug discrimination much faster than nonhuman subjects (Bolin et al., 2016). It is also worth noting that whereas nonhuman animal studies continue to use drug discrimination methods (Porter et al., 2018), drug-discrimination methods are much less commonly used in humans, perhaps due to questions about the clinical utility of this approach (i.e., unlike drug self-administration, there is no direct clinical correlate tied to drug discrimination outcomes; McMahon, 2015, but see Bolin et al., 2016).

Subjective Effects

Subjective drug effects questionnaires capitalize on humans’ unique ability to verbally report their perceptions and responses following drug administration. A diverse range of subjective drug-effects questionnaires has been developed, including measures that directly ask about the effects of a drug (e.g., “How high is the drug making you feel right now?”; Rush et al., 2003), ask about a drug-related effect such as craving or urges to use a drug (e.g., Dudish-Poulsen & Hatsukami, 1997; Tiffany & Drobes, 1991), or that ask about general mood (e.g., Profile of Mood States, see McNair et al., 1971) or the Addiction Research Center Inventory (Haertzen et al., 1963). Regardless of the measure used, responses on these measures are always quantified in some way, for example on a 100-unit visual analog scale, a Likert-type scale, or through summing yes/no or true/false responses to yield a composite score. Like drug discrimination tasks, subjective effects measures are typically administered at regular intervals after drug dosing and can be analyzed as a time course of effects over a session, as a peak effect, area under the curve, or some other summary measure (e.g., a change score).

The subjective responses to drugs of abuse administered in the laboratory tend to follow a characteristic time course, with effects beginning relatively briefly after administration, peaking and then offsetting. The time course of drug effects is influenced by the route of administration (i.e., smoked or intravenously administered drugs tend to have a faster onset and offset than intranasally or orally administered drugs; see Cone, 1995, for data on cocaine administered via multiple routes) and pharmacokinetics of the drug (e.g., drugs with longer half-lives like methamphetamine produce a longer time course of effects than a drug like cocaine which has a shorter half-life; Newton et al., 2005). The magnitude and nature of subjective response observed can be tied to the dose, as well as the efficacy and potency, of the drug administered. For example, Walsh et al. (1995) evaluated the subjective effects of oral methadone (15, 30, and 60 mg) and sublingual buprenorphine (2, 4, and 8 mg). Methadone, a full mu opioid receptor agonist, produced dose-related increases in subjective effects whereas buprenorphine, a partial mu opioid receptor agonist, did not. Like self-administration and drug discrimination approaches, subjective drug effect questionnaires can be used to directly compare the effects of different drugs or the effects of drug combinations (e.g., Carter et al., 2007; Lile et al., 2020; Stoops et al., 2004; Wachtel et al., 2002).

Subjective drug effects are perhaps the most common and simplest “classic” human behavioral pharmacology measure to use because they rely on humans’ long verbal conditioning and ability to describe their interoceptive state unlike the instructions needed for more complex measures like drug discrimination tasks (see further discussion in Schuster & Johanson, 1988). They can also be programmed and administered easily, with relatively few experimental sessions needed to gather data on a range of drugs and doses (i.e., each drug condition to be tested only needs one session for gathering these data). Subjective responses also provide key abuse liability data to regulatory agencies as they are considering scheduling and control of novel psychoactive drugs (Schoedel & Sellers, 2008; Vocci, 1991). Although they provide valid information about a drug’s abuse potential, subjective effects measures may have limited utility in determining whether a putative pharmacotherapy will be effective for treating a substance-use disorder (Comer, Ashworth, et al., 2008a). Subjective responses co-vary with self-administration and drug-discrimination outcomes but they are not isomorphic (Bolin et al., 2013; Reynolds et al., 2013). It is fortunate that subjective effects measures can be used in concert with self-administration or drug-discrimination measures and can provide a more thorough assessment of the human behavioral pharmacology of a given drug or drug combination.

Novel Methods

Human behavioral pharmacology research has expanded beyond these classic measures to evaluate a wide array of other behaviors that can be affected by centrally acting drugs and/or are related to substance-use disorders. With some notable exceptions (e.g., testing the effects of alcohol on craving or of d-amphetamine on impulsive behaviors; see de Wit, 2000, and Strzelecki et al., in press, respectively, for reviews), much of this work has focused on evaluating the target behaviors in the absence of drug administration to compare people who use drugs to controls or to understand how drug cues or other cues affect the outcomes. Unlike the above measures, which cannot be measured in the absence of drug administration, these outcomes can be measured without needing to test direct drug effects. Readers are referred to several outstanding reviews covering the state of the literature as it relates to these types of outcomes, including attentional bias, drug purchasing tasks, delay discounting, and impulsivity (Anselme & Robinson, 2020; Aston & Cassidy, 2019; Bickel et al., 2019; Kozak et al., 2019; Strickland et al., 2020; see also Strickland & Johnson, 2021). This perspective will briefly touch on two measures that have been examined in our laboratory following controlled drug administration at the University of Kentucky: drug purchasing tasks and attentional bias tasks.

Drug purchasing tasks have been modified from the broader behavioral economic literature to determine whether people will make purchases (often hypothetical) of a drug across a range of prices, which yield a range of parameters, including demand intensity, elasticity, breakpoint, and maximum expenditure (Kaplan et al., 2018; Strickland et al., 2020). In our laboratory, we have evaluated the utility of drug purchase tasks as a potential substitute for drug self-administration outcomes, which could be particularly valuable for screening interventions in individuals who cannot safely be administered drugs in the human laboratory (Moeller & Stoops, 2015). In one study, for example, we evaluated the effects of oral placebo and phendimetrazine maintenance (210 mg/day) on intranasal cocaine (0, 20, 40, and 80 mg) self-administration, as well as hypothetical cocaine purchasing (Stoops et al., 2019). Phendimetrazine did not change cocaine self-administration or cocaine purchasing, indicating negative predictive validity; however, more work is needed to determine the positive predictive validity of purchase tasks for screening interventions (but see MacKillop et al., 2019; McClure et al., 2013).

Attentional bias refers to the disproportionate allocation of attention to a drug or drug cue relative to a neutral cue, often measured using eye tracking technology or a visual probe task. People who use drugs routinely display attentional bias to cues for their preferred drugs (Alcorn III et al., 2019; Marks et al., 2014, 2015a; Weafer & Fillmore, 2015). Although the pharmacological mechanisms underlying attentional bias remain to be determined (Luijten et al., 2014) and its clinical relevance questioned (Field et al., 2014, but see Heitmann et al., 2018), the robustness and commonality of this behavior across substance-use disorders, as well as its potential relationship to sign and goal tracking in nonhuman animals (Anselme & Robinson, 2020) indicate its importance for study in the human laboratory, in particular following controlled drug administration. We have evaluated attentional bias following administration of a range of drugs in our laboratory, primarily focusing on how drugs can affect cocaine attentional bias (Alcorn et al., 2020; Bolin et al., 2017; Marks et al., 2015b). In those studies, we found that treatment with putative cocaine pharmacotherapies like methylphenidate or n-acetylcysteine can attenuate attentional bias, whereas alcohol dosing did not change attentional bias.

Methods Summary

Classic human behavioral pharmacology measures have advanced our understanding of drugs of abuse and substance-use disorders. These measures yield differing and complementary information, showing that most drugs of abuse maintain responding on self-administration measures, function as discriminative stimuli, and produce a comparable constellation of positive subjective effects (e.g., drug liking) and indicators of drug strength (e.g., reports of any drug effect). Novel measures are being introduced into human behavioral pharmacology research that test the effects of controlled doses of centrally acting drugs to better understand the pharmacological underpinnings of a range of other substance-use relevant behaviors like drug purchasing, attentional bias, delay discounting, and impulsivity. Each of these methods has strengths and weaknesses, some of which are outlined above, so investigators are advised to consider those, as well as the research questions to be asked, when they are selecting among these methods for any given experiment. The selected combination of human laboratory measures included in any given study can answer a number of research questions (e.g., about the influence of the manipulation on the reinforcing or subjective effects of a drug; about the safety and tolerability of combined drugs), but other approaches are necessary to answer other research questions (e.g., about the efficacy of an intervention to reduce or eliminate drug use in the natural environment, which would require a large clinical trial).

Design Considerations

Designing a rigorous human laboratory behavioral pharmacology study requires substantial planning and consideration. Some key study design elements are outlined here.

Subject Selection

Subjects enrolled in any given human behavioral pharmacology study should have some level of experience with the drug/drugs (or drug class), as well as the routes of administration, under study. For example, if a study plans to evaluate the effects of smoked cannabis, then endorsement of cannabis smoking with some form of collateral verification (e.g., a THC positive urine screen) should be an inclusion criterion for the study. In most cases, subjects should not be exposed to a more invasive route of administration than they have previous experience with (e.g., a subject who has only consumed a drug orally should not receive that drug intravenously in an experiment), although exceptions have been made  based on the similarity of the safety and effects of certain routes (e.g., individuals who smoke cocaine have been given intravenous cocaine experimentally). There are some instances in which it may be necessary to enroll relatively drug naïve subjects, in particular in research seeking to understand initial vulnerability factors. In such cases, it is critical that investigative teams use the least invasive route possible (e.g., oral administration) and they may want to ensure that enrolled subjects have at least some experience with the drug/drug class under study (e.g., if the study is testing the effects of nicotine, then it may be important to determine if enrolled subjects have used nicotine during their lifetime). Further, appropriate follow-up must be conducted to determine whether the experimental exposure has changed the drug-use patterns of individual subjects. It is also important that enrolled subjects are healthy enough to participate in the research safely, so investigative teams need to conduct appropriate mental and physical health screening to rule out any subjects with contraindications to the drugs under study.

Subject Payment

Participation in human behavioral pharmacology research requires a substantive time commitment from research subjects. Screening can require several multihour visits and sessions are routinely 4–8 hr long. Thus, it is customary to compensate subjects for their time. Several approaches can be taken to determine appropriate payments, including a wage payment or reimbursement approach (Dickert & Grady, 1999; Phillips, 2011). In our laboratory we have relied on a wage payment approach and have generally matched payments to the approximate living wage in the state. We routinely structure payments to encourage completion such that subjects receive a payment after each session but also receive a “completion bonus” when finishing the study. Some have raised concerns that paying cash to individuals with substance-use disorders will lead to increased drug use. Although this argument is viewed as somewhat paternalistic, research shows that subjects may use some study payments to purchase drugs but their drug-use behavior does not change after study participation (Festinger & Dugosh, 2012; Thurstone et al., 2010).

Outcome Selection

As noted above, there are numerous outcomes that a research team can choose from in a human behavioral pharmacology study. Because these studies represent a substantial investment in each research subject, it is common for multiple outcome measures to be selected. Measures should be chosen based on study hypotheses and research aims but practical concerns like time course of drug effects, study timeline, cost, and subject burden should also factor into selecting outcomes.

Dose Range and Route of Administration

To understand the effects of the drugs under study more fully, it is advisable to test multiple active doses. Unlike preclinical studies, however, it is difficult to safely test a broad range of active doses in humans. A common approach, instead of giving a log scale range of doses as is done in animals, is to test anywhere between a 2- and 10- fold increase in doses, with higher doses being carefully chosen based on extant safety data and FDA guidelines. In our laboratory, we have administered up to a 10-fold range of intravenous cocaine doses, 3, 10, and 30 mg/70 kg (e.g., Lile et al., 2016) but have also tested smaller dose ranges (e.g., 8, 16, and 24 mg of d-amphetamine; Stoops et al., 2004).

The route of administration to be tested is dependent on the research question, as well as the medical capacity of the research facility. For example, if a research team wanted to understand the effects of alcohol on driving simulator performance, oral alcohol dosing would be sufficient. However, if a team desired to understand the pharmacokinetics and pharmacodynamics of parenterally administered opioids, then intravenous or other nonoral dosing would need to be used. Oral dose administration can be conducted relatively easily in outpatient laboratory facilities if they have reasonable access to medical response in the case of an adverse event. However, drug administration by parenteral routes (e.g., intranasal, intravenous, or smoked [with the exception of smoked tobacco or cannabis]) routinely need to be conducted within a hospital or medically equipped research unit with supervision by physicians and nurses.

A final consideration regarding dosing and route of administration is whether a study team needs to apply for an investigational new drug (IND) with the FDA (note: this is only relevant to investigators conducting research in the United States). Study teams are strongly encouraged to discuss this requirement both with their local institutional review board (IRB) and the FDA but it is the author’s experience that an IND is needed if (1) the study is testing a dose higher than the daily maximum recommended dose for the drug, (2) if a route of administration other than the FDA approved route is being used, (3) a novel, non-FDA approved drug is being tested, (4) the collected data will be used to apply for an FDA indication or affect product marketing and/or (5) the study is testing a drug in U.S. Drug Enforcement Administration Schedule I.

Comparator Drugs

Human behavioral pharmacology studies will often include comparator drugs (i.e., drugs that are not the substance of primary interest). This can be an important design feature for multiple reasons. First, as described above in the section on “Drug Discrimination,” testing an active, negative control condition can distinguish between general drug effects and more specific drug class effects in study outcomes. Second, and in particular for studies of abuse liability, inclusion of a comparator drug with effects that have been well characterized is especially important when testing the effects of a novel psychoactive medication. If the effects of the novel medication are like those of the known comparator, then it can be inferred that it would likely have similar abuse potential. In such studies, a drug with an indication like that for which the novel drug is being developed (e.g., for a pain medication, an opioid would be tested) or that acts in the same receptor system (e.g., for a GABA agonist, a benzodiazepine would be tested) is recommended as the comparator drug.

Placebo Control and Blinding

Understanding that expectancy effects can influence drug response (e.g., Christiansen et al., 2017; Looby & Earleywine, 2009), it is critical that human laboratory studies use placebo controls. The most desirable approach is the use of a double-blind placebo control in which neither the subject nor the research staff interacting with the subject know the drug condition being administered in any given session. Placebos should physically resemble the active doses as closely as possible, for example, in the size, shape, and color of the capsules administered or the color and amount of liquid in an intravenous syringe. If multiple routes of administration or multiple dosing times are used in a session, then a double-dummy approach should also be used. In particular, if both oral and intravenous doses are given in a session, but only one is active, the other dose should still be administered as a placebo. Likewise, if doses are separated by time, but only one is active, the other dose should still be administered as a placebo. Use of blinding and placebo controls can present obstacles to fully informed consent, however, which will be discussed in more detail below.

Within- versus Between-Subjects Designs

Human behavioral pharmacology research can employ either a within- or between-subjects design, based on the research question and investigator preferences. These studies require substantial investment in each subject (e.g., screening costs, training, staff time) so a within-subjects design may be preferable to minimize the number of subjects enrolled while maximizing statistical power. If a within-subjects approach is too burdensome on a subject, however, the likelihood of attrition increases. It is fortunate that modern statistical approaches can manage data missing at random better than the repeated measures ANOVA that has been commonly used to date in this research. But if a study appears to pose a heavy burden when designed using repeated measures, it is still advisable to select a between-subjects approach at the outset. When determining group differences, when a within-subjects design would be too burdensome (e.g., too many sessions to test all experimental conditions) or when drug carry-over effects prevent the use of a within-subjects design, a between-subjects design may be necessary but it is critical that investigators consider how best to match their groups if selecting this approach.

Ethics

The ethics of human behavioral pharmacology research is often questioned, typically by individuals unfamiliar with the field. However, to quote the late luminary, Dr. Marian Fischman, “This research is badly needed. Not to do it would be unethical” (Kleber, 2002, p. 557). Of course, human behavioral pharmacology studies must meet accepted ethical standards, as outlined in more detail below according to the Belmont principles of beneficence, respect for persons, and justice, and undergo rigorous IRB (and other oversight body) review to be conducted safely and ethically. This work must also consider research subjects’ opinions, perceptions, and attitudes about the risks of participation and compensation for engaging in these studies (Strickland & Stoops, 2015).

Beneficence

All human subjects research is expected to produce some form of benefit either directly for the research subject, society, or both. By the same token, the risks to which subjects are exposed must not outweigh the benefits of the research. Thus, human behavioral pharmacology researchers must engage in a careful risk/benefit analysis at the outset of any study and must continue to monitor whether this ratio changes as the research progresses. Some benefits of research available to subjects include the rigorous medical screening that accompanies the research, as well as the remuneration they receive for volunteering their time (but note, not all IRBs consider payment a benefit). It is also important to note that participation in human behavioral pharmacology research has not resulted in changes in drug-use patterns (Kalapatapu et al., 2012; Kaufman et al., 2000).

Risks to individual subjects include physical/medical risks like potential side effects of the drug under study. Careful screening, conducting the research in a comfortable environment near study physician offices with rescue medication and emergency services available and thoughtful selection of the dose/route can help to mitigate these risks. It is also crucial that human behavioral pharmacology experiments establish conservative criteria for terminating dosing and/or participation (e.g., if a subject experiences a heart rate over X beats per minute after a cocaine dose). Participation in human behavioral pharmacology can also pose a threat to privacy and/or confidentiality because a substantial amount of information is collected about subjects’ health and engagement in potentially illegal behavior. It is especially important to consider whether/how an electronic health record at researchers’ academic institution may expose this information to entities outside the research team. Adherence to HIPAA and certificate of confidentiality guidelines help to minimize these risks, in particular because these guidelines limit what health information can be shared with outside entities and, in the case of certificates of confidentiality, afford the ability to resist law enforcement subpoenas for sensitive information about potentially illegal activities.

Societal benefits of human behavioral pharmacology research may be less tangible but are certainly as impactful. They include enhancing understanding of the basic behavioral and pharmacological effects of drugs, aiding in the development of intervention and prevention efforts in substance-use disorders, and potentially influencing policy decisions. A notable example of this was the use of data indicating the behavioral pharmacological similarities between crack and powder cocaine (e.g., both crack and powder cocaine produce rapid and robust positive effects when smoked and injected, respectively; see Hatsukami & Fischman, 1996) to challenge unfair and racially biased sentencing minimums for crack cocaine possession relative to powder cocaine possession in the Fair Sentencing Act of 2010. So, referring to Dr. Fischman’s quote above, without human behavioral pharmacology research to provide data on the clinical behavioral pharmacology of crack and powder cocaine, we would be left to make unscientific assumptions about these two cocaine formulations and striking sentencing disparities could still be in place.

Respect for Persons

The respect for persons principle indicates that individuals must have the autonomy to make decisions for themselves regarding research participation and that individuals who may have limited autonomy receive extra protections. Adherence to this principle is largely achieved through the informed consent process and document. It is critical to recognize that informed consent is an ongoing process throughout study participation and although the consent document is a vital piece of that process, obtaining a subject’s signature on the document is not sufficient to indicate an appropriately conducted informed consent process.

In human behavioral pharmacology research, several key actions are recommended to ensure subjects’ autonomy is protected: (1) subjects’ capacity to consent should be confirmed (e.g., through passing a quiz about the study, through a teach-back method in which subjects explain the study requirements to team members after reading the consent document, through screening methods like a mental status examination); (2) research teams should allow adequate time for subjects to read and process the consent document; and (3) allow subjects to discuss the study, their questions, and concerns with the principal investigator or their delegate.

Human behavioral pharmacology studies often include a blind (see above), which can threaten informed consent. Research teams have navigated this issue in different ways, but in our laboratory, we provide a full list of the drugs/drug combinations/routes of administration to be administered in the study, along with known side effects, in the consent document along with an advisory that on any given day, subjects will not know which drugs/doses they will be receiving. We also offer full debriefing to subjects after they complete the study. Other threats to this principle include (1) overwhelming levels of information in the consent document, which can be difficult to avoid considering all the federally and institutionally mandated text that must be included, but which can be mitigated through use of easy-to-understand language (e.g., written at an eighth-grade reading level as is recommended for most human subjects research); or (2) limited autonomy (e.g., individuals who may be subject to coercion, like prisoners; individuals without capacity to consent, like children) although such vulnerable populations are typically excluded from human behavioral pharmacology studies. As noted above, much early research was conducted with prisoners but this practice was ended in the 1970s following recommendations of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (Branson, 1977).

Justice

Populations who will benefit from human subjects research are expected to participate in that research and research burdens should be distributed across the populations that benefit from it. In action, the principle of justice can perhaps be the most difficult to achieve given the inclusion/exclusion criteria often in place in human behavioral pharmacology research. However, study teams are encouraged to adhere to this principle as much as possible through: (1) thinking broadly about the populations affected by their targeted condition; (2) using as open inclusion/exclusion criteria as safety allows and judiciously applying these criteria; (3) developing plans to successfully recruit diverse populations that match the demographics of their affected populations; and (4) partnering with surrounding communities to ensure respectful and appropriate engagement and recruitment strategies (and, if so desired, providing a report to those communities about the study findings).

Balancing Ethical Principles

No single Belmont principle is paramount and investigative teams must balance them against one another when designing and conducting their research. In practice, this can be quite difficult in human behavioral pharmacology research because of the need to also manage safety concerns. For example, individuals who are at increased risk of adverse events (e.g., someone who has had a myocardial infarction) or who are seeking treatment are routinely excluded from these studies even though these groups are likely to benefit from the findings of the work. As such, the principles of beneficence and respect for persons may come to the forefront in deliberations about the ethics of this research, but investigative teams must also consider justice and determine the ways in which they can attempt to address fair distribution of research burdens. The field has specific standards (see College on Problems of Drug Dependence, 1995; Fischman & Johanson, 1998) that can guide investigators, but it is important to recognize that the field is also evolving (e.g., Roberts et al., 2021, which raises the issue of enrolling treatment seeking individuals in alcohol administration research). When questions or new issues arise, teams are encouraged to speak to their colleagues and/or their IRBs to best navigate them.

Summary and Conclusions

Human behavioral pharmacology research has long been used to rigorously evaluate the effects of centrally acting drugs in a controlled fashion. The work has evolved from evaluating behaviors that map directly to those studied in nonhuman animals to include more complex outcomes to better inform our understanding of drugs of abuse and substance-use disorders. Further, such human laboratory research serves as a translational bridge between nonhuman animal work and clinical trials in addiction. During the planning and execution of this research, important design and ethical considerations must be considered and addressed, leading to the possibility that human behavioral pharmacology studies can become particularly work and resource intensive. Although the resource intensive nature may limit the ability to do some human behavioral pharmacology research to academic medical center settings (e.g., assessment of the pharmacokinetic and pharmacodynamic effects of intravenous cocaine), many studies can be conducted safely and rigorously in other settings if approached creatively by investigative teams. Examples of such studies are those evaluating the behavioral pharmacology of commonly used drugs like alcohol, caffeine, and nicotine, which can often safely be administered through relatively noninvasive routes on an outpatient basis. The dividends paid by these studies include a more thorough understanding of the pharmacodynamic effects of drugs of abuse, advancement of novel substance-use disorder treatments, and development of policy based on science rather than supposition.

Declarations

Conflicts of interest

No relevant conflicts of interest to declare.